Resolver-to-Digital Module
Annotation:
Resolver technology has been used in the industrial environment for many
years. This technology still provides the best angular position transducer
available in terms of ruggedness, reliability and resolution. Re−
solver/Digital Module provides excitation to the resolver and translates
the returned angular analogue information into a digital form for motion
control applications using motors with resolver feedback. The digital an−
gular position output information is available in two forms: As a serial bi−
nary output, providing the absolute position, and as two quadrature sig−
nals A and B, and the Z (zero) pulse, emulating an incremental encoder.
Author:
Petr Kadaník
page 2/17
Resolver−to−Digital Module
Content
INTRODUCTION
3
WHAT IS A RESOLVER?
3
RESOLVER SIGNAL FORMAT
4
BRUSHLESS RESOLVERS
5
RESOLVER-TO-DIGITAL CONVERSION
6
TECHNICAL SUMMARY
8
KEY FEATURES
8
POWER SUPPLY
8
R/D INTEGRATED CIRCUITS
9
AD2S90: R/D CONVERTER
10
AD2S99: PROGRAMMABLE OSCILLATOR
11
OSCILLATOR OUTPUT STAGE
12
I/O CONNECTORS
12
CONCLUSIONS
16
REFERENCES
17
APPENDIX
17
Gerstner Lab  Prague, Jan−2003
Resolver−to−Digital Module
page 3/17
Introduction
Flexible motor control is unthinkable without precise information about the position of the shaft.
For this purpose, different types of shaft angle sensors are used, often built into the driving mo−
tors. On the basis of their physical design, these angular transducers can be classified into two
main groups:
• Optical: A phototransistor or other light−sensitive electronic device counts lines on
a transparent disk mounted to the rotating shaft. The most common of these devices are
incremental and absolute encoders.
• Inductive: Built like small electrical motors, where inductive coupling between a rotating
part (the rotor) and a stationary part (the stator) generates signals indicating shaft position.
Resolvers and synchros are the most common devices.
Optical transducers, especially incremental encoders, have found wide application because their
digital outputs can be easily processed by both discrete logic and microprocessors. Nevertheless,
optical transducers have a number of characteristics that make them less than optimal for many
applications. The built−in semiconductors used to amplify and format the digital output signals are
sensitive to temperature and the LED light sources commonly employed are susceptible to aging.
Furthermore, applications that require an absolute output signal require absolute encoders, which
are much more complicated and therefore expensive.
From a purely practical standpoint, the precise concentricity between the encoder disk and the
sensors required to maintain accuracy as well as the mere presence of optical devices in
an industrial environment dictate that a fully enclosed device with bearings and shaft be used in
all but the crudest applications. Since encoders are typically connected to a shaft having its own
bearings, the user must pay for the second set of high−quality bearings in the transducer as well
as a flexible coupling to connect the two shafts. In many applications, especially brushless servo−
motor commutation or flux vector control of AC induction motors, the additional length of the
optical encoder’s shaft, bearings, and coupling is too great and the optical encoder cannot be
used.
On the other hand, inductive transducers such as resolvers are intrinsically absolute and require
no semiconductors on the transducer itself − the raw output signal can be transmitted over dis−
tances of more than 100 meters. In addition, since they consist primarily of copper and steel, re−
solvers are virtually insensitive to temperature over a wide range. Because no sensitive electronics
or optics are employed, resolvers are often supplied in an unhoused (also called frameless or
pancake) configuration and can be mounted directly to the shaft whose position is to be mea−
sured. Cost and length savings are realized by the user since no shaft−to−shaft coupling or extra
bearings are required.
What is a resolver?
A resolver is a position sensor or transducer which measures the instantaneous angular position of
the rotating shaft to which it is attached. Resolvers and their close cousins, synchros, have been in
use since before World War II in military applications such as measuring and controlling the angle
of gun turrets on tanks and warships. Resolvers are typically built like small motors with a rotor
Gerstner Lab  Prague, Jan−2003
page 4/17
Resolver−to−Digital Module
(attached to the shaft whose position is to be measured), and a stator (stationary part) which pro−
duces the output signals.
The word resolver is a generic term for such devices derived from the fact that at their most basic
level they operate by resolving the mechanical angle of their rotor into its orthogonal or Cartesian
(X and Y) components. From a geometric perspective, the relationship between the rotor angle (θ)
and its X and Y components is that of a right triangle:
Fig.1: Resolving an Angle into its Components
Fundamentally, then, all resolvers produce signals proportional to the sine and cosine of their
rotor angle. Since every angle has a unique combination of sine and cosine values, a resolver
provides absolute position information within one revolution (360°) of its rotor. This absolute
(as opposed to incremental) position capability is one of the resolver’s main advantages over in−
cremental encoders.
Electrically, a traditional resolver, is a transformer in which the coupling between the primary and
the secondaries varies as the sine and cosine of the rotor angle. A resolver has its primary on the
rotor and its secondaries in the stator (necessitating brushes and slip rings or a rotating trans−
former to couple signals into the primary).
Like all transformers, a resolver requires an AC carrier or reference signal (sometimes also called
the excitation) to be applied to its primary. The amplitude of this reference signal is then modu−
lated by the sine and cosine of the rotor angle to produce the output signals on the two seconda−
ries.
For best performance, the reference signal should be a sine wave. In any transformer, there is
a value which relates the output voltage produced by the secondary to that fed into the primary.
For resolvers, this quantity is called the transformation ratio or TR and is specified at the point of
maximum coupling between primary and secondary. For industrial resolvers, the defacto standard
transformation ratio is 0.5, which means that the maximum voltage produced by either secondary
is half the amplitude of the reference signal.
If we define the reference voltage VR , then the voltages on the secondaries are given by the follo−
wing equations:
Sine Secondary: VS = VR TR sinθ
Cosine Secondary: VC = VR TR cosθ
where θ is the mechanical angle of the rotor.
Resolver Signal Format
If we excite the primary (VR) with the recommended sinusoidal reference signal as shown below,
the secondary voltages are also sinusoidal at the same frequency and nominally in phase with the
reference. Their amplitude is proportional to the amplitude of the reference, the transformation
ratio TR, and the sine or cosine of the mechanical angle of the rotor.
While it is helpful to know how the resolver signals appear as functions of time since that is what
one sees when one looks at them with an oscilloscope, it is often more convenient to work with the
Gerstner Lab  Prague, Jan−2003
page 5/17
Resolver−to−Digital Module
envelope (amplitude at the reference frequency) of the signals with respect to rotor position. If the
rotor of the resolver is turned at a rate such that it makes one complete revolution in the time of
10 cycles of the excitation frequency (speeds this high are very difficult and cause other problems
in practice, but are useful for understanding), the envelope of the secondary signals can be clearly
seen. Shown below (Fig.2) is the envelope of the sine secondary signal with respect to rotor posi−
tion.
Fig.2: Modulation Envelope of Sine Secondary Signal
The process of removing the carrier signal − leaving just the envelope − is called demodulation
and is performed by the Resolver−to−Digital (R/D) converter. The demodulated sine and cosine
resolver signals are shown in Fig.3.
Fig.3: Demodulated Resolver Secondary Signals
Brushless Resolvers
The traditional brushless resolver consists of a wound rotor and stator as shown below (Fig.4).
The windings on the rotor generate an AC magnetic field with a sinusoidal distribution. This field
induces voltages in the two stator windings whose amplitudes are dependent on the rotational
angle of the rotor. To provide sine and cosine signals, the two secondaries are wound in space
quadrature (90 physical degrees apart) in the stator. Electrical energy has to be supplied to the
rotor to generate its AC magnetic field. However, as the rotor must be able to rotate freely it is not
possible to use wires. The use of slip rings is also not recommended because they are subject to
wear, generate signal noise, and compromise the mechanical ruggedness of the resolver.
The brushless resolvers therefore use a rotary coupling transformer to transfer energy from the
stator to the rotor. The primary of this rotary transformer is built into the stator. The secondary is
mounted on the rotor and connected directly to the resolver primary. Because of the energy lost in
energizing this two−stage transformer (basically two transformers in series), many turns of wire are
required to generate usable output signal amplitudes. The large number of turns means that the
Gerstner Lab  Prague, Jan−2003
page 6/17
Resolver−to−Digital Module
resolver is a relatively high−impedance device, limiting its use at high excitation frequencies or
rotational speeds.
Because the resolver has a wound rotor, its maximum speed is limited since the windings tend to
fly out of the rotor due to centrifugal force. Typical maximum speeds are 10,000 RPM or less.
Fig.4: Traditional brushless resolver with wound rotor and rotary transformer
Resolver−to−Digital Conversion
While resolvers were originally developed for military and aerospace applications, in recent years
industrial automation has shown more interest in these rugged and precise absolute position
transducers. Nevertheless, the expansion of the use of resolvers was often limited by the fact that
the signal conversion required cumbersome circuitry and automated resolver production was diffi−
cult. Now, however, inexpensive and easy−to−implement monolithic ICs that perform complete
resolver−to−digital conversion are available. These R/D converters give an absolute or incremental
output with a resolution of up to 65,536 counts per revolution. A typical two−chip solution is
shown in Fig.5.
Fig.5: Simple R/D conversion using two monolythic ICs
The resolver signals are low bandwidth amplitude−modulated sine waves. Since these sine wave
signals contain only a single frequency component rather than the virtually infinite frequency spec−
trum of an optical encoder’s square wave signals, they are inherently much more immune to the
high−frequency noise generated by PWM motor drives and other industrial machinery.
Gerstner Lab  Prague, Jan−2003
page 7/17
Resolver−to−Digital Module
The resolver−to−digital converter performs two basic functions: demodulation of the resolver for−
mat signals to remove the carrier, and angle determination to provide a digital representation of
the rotor angle. The most popular method of performing these functions is called ratiometric
tracking conversion. Since the resolver secondary signals represent the sine and cosine of the ro−
tor angle, the ratio of the signal amplitudes is the tangent of the rotor angle. Thus the rotor angle,
θ, is the arc tangent of the sine signal divided by the cosine signal:
θ = arctan
V
sinθ
= arctan S
VC
cosθ
The ratiometric tracking converter performs an implicit arc tangent calculation on the ratio of the
resolver signals by forcing a counter to track the position of the resolver. This implicit arc tangent
calculation is based on the trigonometric identity sin(θ − δ ) ≡ sinθ cos δ − cosθ sinδ .
This equation says that the sine of the difference between two angles can be calculated by cross
multiplying the sine and cosine of the two angles and subtracting the results. Further, as long as
the difference between the two angles is relatively small (δ=θ±30°), the approximation
sin(θ − δ ) ≅ θ − δ may also be used, further simplifying the equation. Thus, if the two angles are
within 30° of each other, the difference between the angles can be calculated using the cross mul−
tiplication shown above. In the R/D converter, this equation is implemented using multiplying D/A
converters to multiply the resolver signals (proportional to sinθ and cosθ) by the cosine and sine of
the digital angle, δ, which is the output of the converter.
The results are subtracted, demodulated by multiplying by the reference signal, and filtered to give
a DC signal proportional to the difference or error between the resolver angle, θ, and the digital
angle, δ. The digital angle, δ, stored in the counter, is then incremented or decremented using a
voltage controlled oscillator until this error is zero, at which point δ=θ (the digital angle output of
the converter is equal to the resolver angle). This incrementing and decrementing of the digital
angle, δ, causes it to track the resolver angle, θ, hence the name of this type of converter.
Gerstner Lab  Prague, Jan−2003
page 8/17
Resolver−to−Digital Module
Technical Summary
Resolver−to−digital interface module offers an effective solution for motion control applications
using motors with resolver feedback. It provides the resolver excitation and translates the returned
angular analogue information into a digital form.
The digital angular position output information is available in two forms:
• As a serial binary output, providing the absolute position. This output is available on
the serial peripheral connector (SPI) pins. The absolute position has 12 bits resolution pro−
viding 4096 values per rotation.
• As 2 quadrature signals A and B, and a Z (zero) pulse, emulating an incremental en−
coder. These signals are available on the quadrature encoder connector pins. The encoder
emulated outputs continuously produce signals equivalent to a 1024−line encoder.
Key features
•
•
•
•
•
12−bit serial absolute position
4x1024−lines incremental encoder emulation
Differential inputs for resolver signals
Internal oscillator providing sinusoidal excitation with an arbitrary frequency 2−20kHz
External power supply 7 ÷ 25V DC @ 100mA is required
Power supply
The printed circuit board requires 7 ÷ 25V DC @ 100mA power supply. Input power circuit deli−
vers +5V DC to an isolated DC/DC converter providing bipolar ±5V DC voltage output galvani−
cally isolated from an input power supply. A green LED indicates power applied to the modul.
Schematic diagram of power supply stage is shown in Fig.6. The main power input is either
through connector J1 (2−pin connector) or connector J2 (2.1mm coax power jack). VSS is a net
label for +5V DC voltage and VDD is a net label for –5V DC voltage. GND is a common ground
for both digital and analog signals galvanically isolated from an input power supply (GND_PWR).
Gerstner Lab  Prague, Jan−2003
page 9/17
Resolver−to−Digital Module
Fig.6: Schematic diagram of the Power supply circuit
R/D Integrated Circuits
The core of the R/D modul is created by two monolithic ICs – the reference generator unit
AD2S99 which provides a constant amplitude sinusoidal excitation for the resolver and the 12−bit
R/D converter AD2S90 which translates differential sine and cosine signals generated by the re−
solver into a digital information.
You can see circuit configuration of these two ICs in Figure 7.
Gerstner Lab  Prague, Jan−2003
page 10/17
Resolver−to−Digital Module
Fig.7: R/D Converter (AD2S90) and Oscillator (AD2S99) configuration
AD2S90: R/D Converter
The AD2S90 is a 12−bit resolution monolithic resolver−to−digital converter. It provides a complete
solution for digitizing resolver signals without using external components. To get position angle θ
in digital form, both sine and cosine channel signals must be sent to the R/D converter for decod−
ing. This chip provides not only absolute position data output (in serial form), but also standard A
Quad B format signals. The absolute position data output gives additional useful information. The
serial form of the digital position data is especially suitable for long distance data transmission. All
the input/output signals are TTL logic compatible.
Functional block diagram is shown in Fig.8. Please, see Appendix B for more details about
AD2S90.
Gerstner Lab  Prague, Jan−2003
page 11/17
Resolver−to−Digital Module
Fig.8: Functional block diagram of AD2S90
The AD2S90 emulates a 1024−line encoder. Relating this to converter resolution means one revo−
lution produces 1025 A, B pulses. B leads A for increasing angular rotation. The addition of the
DIR output negates the need for external A and B direction decode logic. DIR is HI for increasing
angular rotation. The north marker NM pulse is generated as the absolute angular position pas−
ses through zero.
AD2S99: Programmable Oscillator
The AD2S99 programmable sinusoidal oscillator provides sine wave excitation for resolvers and
a wide variety of AC transducers. The AD2S99 also provides a synchronous reference output sig−
nal (3 V p−p square wave) that is phase locked to its SIN and COS inputs. In an application, the
SIN and COS inputs are connected to the transducer’s secondary windings. The synchronous ref−
erence output compensates for temperature and cabling dependent phase shifts and eliminates
the need for external preset phase compensation circuits.
The excitation frequency is easily programmed to 2 kHz, 5 kHz, 10 kHz, or 20 kHz by using the
frequency select pins. Intermediate frequencies are available by adding an external resistor.
Fig.9: Functional block diagram of AD2S99
Functional block diagram is shown in Fig.9. Please, see Appendix B for more details about
AD2S99.
Gerstner Lab  Prague, Jan−2003
page 12/17
Resolver−to−Digital Module
Oscillator output stage
The output of the AD2S99 oscillator consists of two sinusoidal signals, EXC, and EXC. EXC is 180°
phase advanced with respect to EXC. With low impedance transducers, it may be necessary to
increase the output current drive of the AD2S99. In such an instance, an external buffer amplifier
can be used to provide gain (as needed), and additional current drive for the excitation output
(either EXC or EXC) of the AD2S99, providing a single ended drive to the transducer. Refer to
Fig.10 for a buffer configurations.
Fig.10: Circuit scheme for an excitation signal gain adjustment
I/O Connectors
Fig.11 presents a top view of the R/D module which outlines the main components and the con−
nectors on printed circuit board (PCB).
Gerstner Lab  Prague, Jan−2003
page 13/17
Resolver−to−Digital Module
Fig.11: R/D Module board layout
The excitation frequency for the resolver can be easy selected by jumpers setting on JP1 header
according to the following table
Frequency Jumper position (pin numbers) on JP1
Selection
12
34
56
78
2kHz
O
I
I
O
5kHz
O
I
O
I
10kHz
I
O
I
O
20kHz
I
O
O
I
Jumper JP2 has to be closed if you don’t want to set up intermediate frequency value. See
AD2S99 DataSheet in Appendix B for more details.
The function of jumpers JP3 and JP4 is understandable from Fig.10. They can bypass operational
amplifiers in excitation output signal buffer circuit. Signal gain is 1 when JP3 and JP4 are closed.
Jumper JP3 is bypassing signal EXC_Hi and jumper JP4 is bypassing signal EXC_Lo.
Output Encoder connector (J4) is shown in Fig.12.
Fig.12: Encoder Output Connector
Output Serial connector (J5) is shown in Fig.13.
Gerstner Lab  Prague, Jan−2003
page 14/17
Resolver−to−Digital Module
Fig.13: Serial Output Connector
Output connector (J6) signed Add is shown in Fig.14. This connector contents signals not used for
standard Quad Encoder or Serial digital communication but are presented in AD2S90 output
pins.
Fig.14: Add Output Connector
The analog velocity output VEL is scaled to produce 150 rps/V DC ±15%. The sense is positive for
increasing angular rotation.
I/O Resolver connector (J3) is shown in Fig.15.
Fig.15: Resolver I/O Connector
The resolver shall be connected to the module according to Fig.16.
Fig.16: Resolver connection to the R/D module
6−wire cable with a common shield was used in our lab for a signal transmission from an auxiliary
BNC connectror box to the R/D module (connector J3). The following table assignment is specific
only for this application.
Gerstner Lab  Prague, Jan−2003
page 15/17
Resolver−to−Digital Module
Signal
Wire Color
SIN_Hi
Blue
SIN_Lo
Grey
COS_Hi
brown
COS_Lo
black
EXC_Hi
red
EXC_Lo
Yellow
Gerstner Lab  Prague, Jan−2003
page 16/17
Resolver−to−Digital Module
Conclusions
The main purpose of this report is to describe the Resolver−to−Digital module. This modul is an
interlink between the resolver (analog measurement of speed and position) and the Digital signal
processor (digital control of frequency converter). Used control board with Motorola DSP56F805
is equipped with a peripheral circuit assigned to process quadrature signals from an incremental
encoder.
DC motor in our lab (used as a variable load) has a resolver integrated on its shaft. The applying
of R/D module was therefore a natural and the easiest option, how to measure motor speed with−
out replacing the resolver by the incremental encoder.
Gerstner Lab  Prague, Jan−2003
page 17/17
Resolver−to−Digital Module
References
•
Resolver−to−digital Interface Module RDIM16, User Manual, Technosoft 2001
•
Circuit Applications of the AD2S90 Resolver−to−Digital Converter, Analog Devices Applica−
tion Note, AN−230
•
Digital Resolver Integration, Analog Devices Application Note, AN−234
•
A New Absolute Inductive Transducer for Brushless Servomotors, Technical Paper, Ad−
motec Inc., 1995
Appendix
All schematics are included in this appendix. For schematics and PCB design was used software
package Protel 99. This appendix includes also datasheets and technical specifications of compo−
nents implemented in R/D module design.
Gerstner Lab  Prague, Jan−2003
3
C10
100nF
1N4007
12V-PWR
+
D2
CT4M7/10V - Tantal
C14
4.7uF
Vdd
+
JP2
+
1000uF/16V
+
Freq.Aprox
100nF
1
10
9
100nF
GND
3
Vss
Vdd
NM
ClkOut
sin LO
EXC_Hi
7
8
2
B
GND
TLV2460
U6
NM
A
4
DIR
Data
1
sin
B
3
Vss
DGnd
A
2
SIN_Lo
AGnd
CS/
SIN_Hi
6
EXC_Hi
5
AD2S90
cos
6
GND
2
Ry
Vdd
1
SCLK
COS_Hi 20
Vdd
5
JP3
cos LO
CS/
1
Vel
GND
COS_Lo 19
R9
R7
1k
C
Vdd
CF1-1M0
Rx
17 Vel
13
SCLK
GNDout
+Vout
6
5
100k
Vss GND Vdd
R5
LOS
GND
CF1-1M0
exc/
U2
AD2S99
GND
11
U5F 74HC04
14
R4
50k
U5E 74HC04
C7
GND
R3
8
12
exc
Vdd
U5D 74HC04
14
SYNCout
-Vout
9
Ref
DCP010505DB
C16
18
GNDin
Vin
8
SYNCin
7
3
U5B 74HC04
U4
GND-PWR
GND
4
17
15
4
2
MKS4-2M2-63V
1
C5
LED
U5A 74HC04
exc/
14
2k5
Vss
nc
+
R2
SEL2
nc
nc
15 NMC
2
100uF/10VM
cos
nc
3
C4
8
18
16
AGnd
Vdd
Vout
GND
C3
DGnd
7
exc
nc
6
GND
COS_Hi
AD2S99
sin
13
5
Vdd
D
exc/
LOS
SIN_Hi
nc
12
C17
C18
C19
100nF 100nF 100nF
GND
4
11
4.7uF
C15
16 ClkOut
100nF
C11
SynRef
4.7uF
C13
SEL1
FBias
100nF
C9
Data
2
GND-PWR
C6
GND
Vss
Vss
U3
78M05
Vin
JACK_PWR
C
Vss
Freq.Selection
10
100nF
1
C2
D3
2
4
6
8
+
D
C1
1
GND
1
3
5
7
R1
Rf
GND
1
2
3
JP1
2
BZX83V008.2
J2
+
sh
Vdd
19
C12
4.7uF
20
C8
100nF
1
D1
2
1
4
3
J1
2
2
1
13
12
11
Vss
GND
10
DIR
9
NM
U1
AD2S90
GND
GND
R8
1k
6
Vdd
5
3
B
exc
1
R6
EXC_Lo
TLV2460
U7
GND
2
Rx
JP4
GND
R10
1
2
3
Ry
B
Vdd=+5V
Vss=-5V (+-5%)
-------------------------------Freq.output selection:
Freq | SEL1 | SEL2
2kHz | Vss | Vss
5kHz | Vss | GND
10kHz | GND | Vss
20kHz | GNH | GND
--------------------------------
4
EXC_Lo
J3
8
7
6
5
4
3
2
1
EXC_Lo
EXC_Hi
COS_Hi
COS_Lo
SIN_Lo
SIN_Hi
GND
GND_AD
GND
GND
Resolver I/O
J4
A
B
NM
DIR
J8
A
8
7
6
5
4
3
2
1
EXC_Lo
EXC_Hi
COS_Hi
COS_Lo
SIN_Lo
SIN_Hi
GND
GND
74HC04
J5
5
CS/
DATA
Encoder OUT
1
2
3
4
GND
J6
LOS
Vel
ClkOut
Serial OUT
1
2
3
4
Add OUT
GND
Title
GND
3
2
1
Size
Vss
Vdd
Resolver-to-Digital Board
Number
Revision
1.0
A4
PWR
1
SCLK6
GND
J7
TestPoints
U5C
1
2
3
4
5
Date:
File:
2
15-Jan-2003
D:\Protel Design\Resolver-to-digital.ddb
3
Sheet of
Drawn By:
Petr Kadaník
4
A